This supplements, rather than replaces, the existing caching in `PDFFunction.constructPostScript` since that one still makes sense given that functions are cached on the page-level.
Using an additional cache helps improve performance because:
- There are many fewer function calls, and overall less parsing this way.
- For the common case of `Uint8Array`-data we're able to use integer cache-keys, which is a lot faster than string concatenation.
This significantly improves performance of the `pr5134` test-case with `isEvalSupported = false` set, and testing locally in the viewer:
- With the `master` branch and `isEvalSupported = true`, page 2 renders in approx. 340 milliseconds.
- With the `master` branch and `isEvalSupported = false`, page 2 renders in approx. 1200 milliseconds.
- With this patch and `isEvalSupported = false`, page 2 renders in approx. 380 milliseconds.
While this is obviously still slower, the difference is now small enough that it shouldn't be too much of an issue in practice (compare with PR 18070) and the `pr5134` test-case is an especially "bad" one w.r.t. its PostScript function use.
f7ec74c474